X-Rays from RU Lupi: Accretion and Winds in Classical T Tauri Stars

A&A 473, 229–238 (2007) Astronomy DOI: 10.1051/0004-6361:20077644 & c ESO 2007 Astrophysics

X-rays from RU Lupi: accretion and winds in classical T Tauri stars

J. Robrade and J. H. M. M. Schmitt

Universität Hamburg, Hamburger Sternwarte, Gojenbergsweg 112, 21029 Hamburg, Germany e-mail: [email protected] Received 13 April 2007 / Accepted 18 June 2007

ABSTRACT

Context. Low-mass stars are known to exhibit strong X-ray emission during their early evolutionary stages. This also applies to classical T Tauri stars (CTTS), whose X-ray emission differs from that of main-sequence stars in a number of aspects. Aims. We study the specific case of RU Lup, a well known accreting and wind-driving CTTS. In comparison with other bright CTTS we study possible signatures of accretion and winds in their X-ray emission. Methods. Using three XMM-Newton observations of RU Lup, we investigate its X-ray properties and their generating mechanisms. High-resolution X-ray spectra of RU Lup and other CTTS are compared to main-sequence stars. We examine the presence of a cool plasma excess and enhanced plasma density in relation to X-rays from accretion shocks and investigate anomalous strong X-ray absorption and its connection to winds or circumstellar material. Results. We find three distinguishable levels of activity among the observations of RU Lup. While no large flares are present, this variability is clearly of magnetic origin due to the corresponding plasma temperatures of around 30 MK; in contrast the cool plasma component at 2–3 MK is quite stable over a month, resulting in a drop of average plasma temperature from 35 MK down to 10 MK. Density analysis with the O vii triplet indicates high densities in the cool plasma, suggesting accretion shocks to be a significant contributor to the soft X-ray emission. No strong overall metal depletion is observed, with Ne being more abundant than Fe, that is at solar value, and especially O. Excess emission at 6.4 keV during the more active phase suggest the presence of iron fluorescence. Additionally RU Lup exhibits an extraordinary strong X-ray absorption, incompatible with estimates obtained at optical and UV wavelengths. Comparing spectra from a sample of main-sequence stars with those of accreting stars we find an excess of cool plasma as evidenced by lower O viii/O vii line ratios in all accreting stars. High density plasma appears to be only present in low-mass CTTS, while accreting stars with intermediate masses (2 M) have lower densities. Conclusions. In all investigated CTTS the characteristics of the cooler X-ray emitting plasma are influenced by the accretion process. We suspect different accretion rates and amounts of funnelling, possibly linked to stellar mass and radius, to be mainly responsible for the different properties of their cool plasma component. The exceptional X-ray absorption in RU Lup and other CTTS is probably related to the accretion flows and an optically transparent wind emanating from the star or the disk. Key words. stars: individual: RU Lupi – stars: pre-main sequence – stars: activity – stars: coronae – X-rays: stars

1. Introduction accreted from the stellar disk onto the star at almost free-fall velocity along magnetic field lines which disrupt the accretion disk T Tauri stars as a class, are very young low-mass stars. The mem- in the vicinity of the corotation radius. Upon impact, a strong bers of the subclass of so-called classical T Tauri stars (CTTS) shock is formed near the stellar surface and the funnelling of the are still accreting matter from a surrounding circumstellar disk. accreted matter by the magnetic field leads to “accretion spots" CTTS are thought to evolve first into weak-line T Tauri stars with filling factors at the percent level with respect to the stellar (WTTS), where they become virtually disk-less and no longer surface; therefore only a small fraction of the stellar surface is show signs of significant accretion, and eventually into solar- covered by the accretion spots. This accretion shock plasma is like star on the main sequence. Ongoing accretion in a CTTS expected to reach temperatures of up to a few MK and thus to is evidenced by an emission line spectrum and specifically, a lose a large fraction of its energy at UV and soft X-ray wave- large Hα equivalent width (EW > 10 Å). Further, the strong ob- lengths. The accreted plasma does not necessarily have to be at served infrared excess indicates the presence of a disk. In con- high density. However, if one wants to achieve the inferred small trast, WTTS have much weaker Hα emission and little or no filling factors at the percent level and the otherwise determined −8 −7 −1 IR-excess. The increasing dominance of the underlying contin- mass accretion rates of 10 –10 M yr , the infalling gas must uum emission over the optical stellar absorption line spectrum then be at rather high density (n > 1011 cm−3). The accretion gives rise to another sub-classification into moderate, veiled and process is accompanied by outflows from the star and possibly extreme T Tauri stars. However, these strict subdivisions are also from the disk, which probably also play an important role somewhat arbitrary and blurred since for example Hα emis- in the transport of angular momentum. Different mass accretion sion is highly variable and differs also within the CTTS class rates and filling factors, different stellar properties such as mass, by more than an order of magnitude. A detailed account of the rotation and activity and different viewing angles, naturally lead pre-main sequence stellar evolution of low-mass stars is given to the observed variety of the different CTTS phenomena. by Feigelson & Montmerle (1999). X-ray emission from T Tauri stars has already been detected In the commonly accepted magnetospheric accretion model with the Einstein and ROSAT observatories, and both types are for CTTS (Calvet & Gullbring 1998) material is assumed to be found to be copious and variable X-ray emitters. The origin of

Article published by EDP Sciences and available at http://www.aanda.org or http://dx.doi.org/10.1051/0004-6361:20077644 230 J. Robrade and J. H. M. M. Schmitt: X-rays from RU Lupi: accretion and winds in classical T Tauri stars the observed X-ray emission is currently the subject of some de- Table 1. Observing log of the XMM-Newton RU Lup exposures. bate. Since all main-sequence “cool stars”, i.e. stars with outer convection zones, are surrounded by X-ray emitting coronae Observation MOS (filt.) OM (Schmitt & Liefke 2004), some kind of magnetic activity is Date Time Dur.(ks) Exp.(no.) also expected for their pre-main sequence counterparts with their August 08 2005 04:19–12:54 30(28) 6 deep outer convection zones. Indeed, large sample studies such August 17 2005 09:17–17:02 27(15) 6 as COUP (Chandra Orion Ultradeep Project) showed that most Sept. 06/07 2005 18:30–02:16 27(22) 6 of their observed X-ray emission is of coronal origin (Preibisch et al. 2005). Further, magnetic fields, connecting star and disk, −1 have been invoked to explain the huge flares observed in CTTS both infalls and outflows with velocities up to 300–400 km s . (Favata et al. 2005a). Specifically, Lamzin (2000) attributes the origin of the observed On the other hand, the presence of accretion streams and lines profiles in the HST UV spectra to accretion shocks and outflows opens up additional possibilities for the generation of stellar winds. X-rays. Collimated winds have been observed as X-ray jets in From optical high-resolution spectra, Stempels & Piskunov ∼ Herbig-Haro objects (Pravdo et al. 2001; Favata et al. 2006) and (2002) determined mass and radius of RU Lup to M 0.8 M ∼ in CTTS such as DG Tau (Güdel et al. 2005). Accretion shocks and R 1.7 R. They further derived a projected rotational ve- = −1 are expected to generate plasma at a significantly higher density locity of v sin(i) 9kms and detected variability on short time scales of 1h. Together with various proposed periods (0.8– than stellar coronal plasma and their soft X-ray emission can ◦ ◦ be traced by density sensitive line ratios, e.g. between forbidden 3.7 d) this suggests an inclination between 3 and 16 ,i.e.a and intercombination lines in the He-like triplets of O vii and system that is viewed nearly pole-on with rather high intrinsic Ne ix. Anomalously low ratios of these lines have been observed rotation. Finally, they derived an interstellar absorption column = × 19 −2 for several CTTS, e.g. TW Hya (Kastner et al. 2002; Stelzer & towards RU Lup with a corresponding N(H) 6.3 10 cm . Schmitt 2004), BP Tau (Schmitt et al. 2005), CR Cha (Robrade Combining literature data with results derived from FUV data / & Schmitt 2006), V4046 Sgr (Günther et al. 2006) and MP Mus obtained with FUSE and HST STIS, Herczeg et al. (2005) es- (Argiroffi et al. 2007), and the inferred high plasma densities timated RU Lup to be an 0.6–0.7 M star with an age of ∼ × −8 suggest the presence of X-ray emission from shocks as postu- 2–3 Myr and a mass accretion rate of 5 10 M. Their de- = ± −2 lated in the magnetospheric accretion scenario. rived hydrogen column density of log N(H) 20.1 0.2cm , ∼ However, not all accreting stars seem to show high density corresponding to an extinction of AV 0.07, is again quite low. plasma. The prototypical, but not necessarily typical T Tauri star, Combining and summarising these findings we conclude that T Tau itself, is an example where density diagnostics of cool RU Lup is probably an optically rather mildly absorbed system plasma as seen in O vii indicate a lower density which is more viewed almost pole-on. comparable to active main sequence stars (Güdel et al. 2007). While the Einstein IPC failed to detect X-ray emission from × −13 −2 −1 Further, accretion is only capable of producing plasma with tem- RU Lup with an upper flux limit of 1.2 10 erg cm s peratures of up to a few MK (e.g. Günther et al. 2007), and in the 0.5–4.0 keV band (Gahm 1980), X-ray emission from therefore contributes nearly exclusively to the low energy X-ray RU Lup was detected in 1993 in a pointed 15.7 ks ROSAT × −3 / emission. However, in all CTTS – even in the accretion domi- PSPC (0.1–2.4keV) observation at a rate of 9 10 cts s nated, “soft” TW Hya – high temperature plasma is observed, (2RXP catalogue). Using the energy conversion factors given by Neuhäuser et al. (1995), this corresponds to an X-ray flux which together with the strong flaring requires magnetic activity −13 −2 −1 of ∼1 × 10 erg cm s and an X-ray luminosity of log LX = to also be present in these objects. A comparative study of sev- −1 eral bright CTTS indicated the presence of X-rays from both ac- 29.4ergs for a distance of 140 pc. cretion and magnetic activity, but with different respective con- In this paper we present three new XMM-Newton observa- tributions in the individual objects (Robrade & Schmitt 2006). tions of RU Lup with a total exposure time of 84 ks, performed Altogether, magnetic activity is the dominant source of the X-ray over the course of a month in 2005. Our paper is structured as emission in most CTTS at least at higher energies, nonetheless follows. In Sect. 2 the X-ray observations and the data analysis accretion can significantly contribute to the soft X-ray emission. are described, in Sect. 3 we present our results derived from the ff RU Lup is a late K-type CTTS with a high accretion rate and XMM-Newton observations of RU Lup subdivided into di erent an extreme optical veiling. It has been extensively studied espe- physical topics. Finally, in Sect. 4 we compare RU Lup to results cially at optical and UV wavelengths and a review of its basic from X-ray diagnostics of several CTTS and specifically discuss stellar properties is given by Lamzin et al. (1996) and Stempels accretion and wind scenarios in the context of their X-ray emis- & Piskunov (2002). Lamzin et al. (1996) present a magneto- sion in general, which is followed by our conclusions in Sect. 5. spheric accretion model for RU Lup based on multi-frequency monitoring and high-resolution optical data. While there was 2. Observations and data analysis some debate about the distance of RU Lup (140–250pc), recent estimates indicate a value of ∼140 pc (Hughes et al. 1993), RU Lup was observed by XMM-Newton three times in the course which we adopt for this paper. Already Lamzin et al. (1996) of a month in August/September 2005 for roughly 30 ks each; pointed out that their derived reddening sets an upper limit to the detailed observation times are listed in Table 1. Data were RU Lup’s distance of ∼150 pc, when attributed only to interstel- taken with all X-ray detectors, the EPIC (European Photon lar material, also indicating very little circumstellar absorption. Imaging Camera), consisting of the MOS and PN detectors and RU Lup is highly variable and shows strong accretion signatures the RGS (Reflection Grating Spectrometer) as well as the opti- with a large Hα EW of 140 Å up to 216 Å (Appenzeller et al. cal monitor (OM). The EPIC instruments were operated in full 1983). Virtually no photospheric absorption lines are detected frame mode with the medium filter, while the OM was oper- in its spectrum due to substantial optical veiling, on the con- ated with the UVW1 filter, which is sensitive in the waveband trary, the optical spectrum shows many emission lines. The line between 2000–4000 Å with an effective wavelength of 2910 Å. profiles are also strongly variable and asymmetric, suggesting The OM exposures typically lasted around 4 ks and allowed an J. Robrade and J. H. M. M. Schmitt: X-rays from RU Lupi: accretion and winds in classical T Tauri stars 231 accurate determination of RU Lup’s UV brightness. During each X-ray observation, six exposures of the central field containing RU Lup were taken. All data analysis was carried out with the XMM-Newton Science Analysis System (SAS) version 7.0 (Loiseau et al. 2006). Standard selection criteria were applied to the data, and periods with enhanced background due to proton flares were dis- carded from spectral analysis utilising the high energy event rates of the respective detectors. All X-ray light curves are background subtracted, the background was taken from nearby source-free regions. Spectral analysis of the EPIC data was carried out with XSPEC version 11.3 (Arnaud 1996) and is performed in the energy range 0.2–10.0keV. The RGS data of the individual observations is of very low SNR, we therefore merged the data from all three observations using the tool rgscombine. We note that the X-ray source RX J1556.4-3749 of unknown ff type lies in the read-out direction of the RGS, the o set in the Fig. 1. Light curves of RU Lup determined with the PN instrument (1 ks dispersed spectra is around –0.5 Å. Therefore RX J1556.4-3749 binning) in the energy band 0.2–5.0 keV (upper panels) and OM UVW1 could in principle, contaminate the RGS spectra of RU Lup, but magnitude per exposure (lower panels). being substantially absorbed, as deduced from its EPIC spectra, its contribution to the spectral lines and especially the O vii triplet discussed here is negligible. To reduce background and confidence range and were calculated separately for abundances detector noise effects we extracted RGS spectra from the stan- and temperatures by allowing variations of normalisations and dard source region at phases of very low background (95% PSF, respective model parameters. Note that additional uncertainties background rate <0.25 cts/s) as well as from the line core at may arise from errors in the atomic data and instrumental cali- low background (66% PSF, background rate <1cts/s) and cross- bration which are not explicitly accounted for. checked the derived results. For line fitting purposes we use the CORA program (Ness & Wichmann 2002), using identical line width and assuming Lorentzian line shapes. Emitted line fluxes 3. Results are corrected for absorption by using the ismtau-tool provided in 3.1. X-ray light curves and variability the PINTofALE software (Kashyap & Drake 2000). The XMM-Newton X-ray light curves of RU Lup are shown in The data of the EPIC detectors are analysed simultaneously the upper panel of Fig. 1. We observe a steady decline in X-ray for each observation but were not co-added. They are modelled brightness by a factor of three over a month. Note that because with a multi-temperature model assuming the emission spec- of the large time gaps between the observations (9 and 20 days trum of a collisionally-ionized, optically-thin gas as calculated respectively) this decline does not need to be monotonic. Indeed, with the APEC code (Smith et al. 2001). We find that a three- already in the first observation, LX declined by roughly 40% temperature model adequately describes the data. Absorption in within 25 ks but started to rise again towards the end of the the circumstellar environment and in the interstellar medium is observation. During the second observation the count rate in- significant and is applied in our modelling as a free parame- creased by roughly 50% and short-term X-ray variability is quite ter. Abundances are modelled relative to solar photospheric val- common in all observations. We can rule out the presence of ues as given by Grevesse & Sauval (1998). The low FIP (First strong flaring within the individual observations, however the Ionisation Potential) elements Al, Ca, Ni are tied to the Fe abun- August 08 light curve may be interpreted as a part of the decline dance; for other elements with no significant features in the phase of a larger flare. At any rate, the observed overall decline measured X-ray spectra, solar abundances are used. Since ab- during the XMM-Newton observations is very likely accidental sorption, the coolest plasma temperature and abundances do not ff due to sparse temporal sampling. significantly di er between the observations, they are modelled The OM UVW1 flux observed from RU Lup is plotted in the with variable, but tied values to ensure a consistent analysis of lower panel of Fig. 1. Note that derived magnitude errors are in the data. We then derived the temperatures and volume emis- the range of 0.01 mag and below the size of the shown symbols. = sion measures (EM nenHdV) of the individual plasma com- We find variations in UVW1 brightness of up to 0.5 mag between ponents for all three observations and calculated corresponding the three observations, however these variations do not correlate X-ray luminosities from the resulting best fit models. Some of with the observed X-ray brightness and have a far lower ampli- the fit parameters are mutually dependent. This interdependence tude compared to the variations seen in the X-rays. Additional ff mainly a ects the strength of absorption and emission measure UVW1 variability is observed within each observation of the or- of the cooler plasma at a few MK, but also components of the der 0.1–0.15 mag. Some of these variations appear to be corre- emission measure distribution (EMD) and abundances of ele- lated with the X-ray flux, possibly in the second and especially ments with emission lines in the respective temperature range. during the third and X-ray faintest observation. Unfortunately ff Consequently, models with di erent absolute values of these pa- a time-resolved X-ray spectral analysis within each observation ff rameters but only marginal di erences in its statistical quality suffers from the poor signal and a correlation for six exposures may be applied to describe the data. might be a chance effect. Nonetheless, the fact that the corre- Our fit procedure is based on χ2 minimization, therefore lation is most strongly present during the low-activity phases, spectra were rebinned to satisfy the statistical demand of a where the soft X-ray component becomes more important, sup- minimum value of 15 counts per spectral bin. All errors de- ports the view that at least some of the X-ray emission might rived in spectral fitting are statistical errors given by their 90% be caused by the same physical process as the observed UV 232 J. Robrade and J. H. M. M. Schmitt: X-rays from RU Lupi: accretion and winds in classical T Tauri stars

We investigated in detail the decline in X-ray brightness throughout the observations and find that there are only marginal changes in the EM of the cool component around 2 MK; in contrast, the EM of the hot component decreases significantly by a factor of roughly six. At its brightest phase the EMD is dominated by a hot plasma component at 30 MK, and its decline in EM is accompanied by a moderate cooling. In the third observation the intermediate plasma component is also significantly weaker and slightly cooler compared to the other two observations. Consequently, the average coronal temperature drops from 25 MK over 15 MK to 11 MK during the campaign, mainly caused by the decline of EM at temperatures around 20–35 MK. These findings suggest that the fading of RU Lup by a factor Fig. 2. X-ray spectra (PN) of RU Lup (crosses) and spectral fit (his- of three in emitted LX is caused by a decline of its magnetic togram) for the three observations; top down: 08/08 (black), 17/08 (red), activity over a month, and hence that lack of any correlation be- 06/09 (blue). tween X-ray and UV brightness for the three observations is not surprising. 21 −2 Table 2. Spectral fit results for RU Lup, units are NH in 10 cm , kT To compare our results with previous X-ray measurements, 52 −3 / 29 −1 in keV, EM in 10 cm and LX observed emitted in 10 erg s . we also calculated our model fluxes in the Einstein (0.5–4.0 keV) and ROSAT (0.1–2.4 keV) bands. We find that RU Lup was ac- Par. 08/08 17/08 06/09 tually in a phase of enhanced activity in autumn 2005, in agree- +0.2 +0.2 +0.2 NH 1.8−0.3 1.8−0.3 1.8−0.3 ment with the results from our spectral fitting. Even during the Fe 1.07+0.13 1.07+0.13 1.07+0.13 third and X-ray darkest XMM-Newton observation, the deter- −0.20 −0.20 −0.20 mined flux of RU Lup is nearly 20% above the ROSAT value +0.29 +0.29 +0.29 Si 0.64−0.27 0.64−0.27 0.64−0.27 and similar to the Einstein upper limit. +0.19 +0.19 +0.19 From our spectral fits we derive a moderate absorption col- O0.60−0.27 0.60−0.27 0.60−0.27 umn of N = 1.8 × 1021 cm−2 when compared to typical val- Ne 1.38+0.35 1.38+0.35 1.38+0.35 H −0.31 −0.31 −0.31 ues for CTTS (Robrade & Schmitt 2007). However, recalling +0.02 +0.03 +0.04 / kT10.18−0.03 0.18−0.03 0.18−0.03 that the optical UV extinction measurements suggest columns ∼ 20 −2 kT20.63+0.02 0.63+0.03 0.57+0.03 of 10 cm and the fact that RU Lup is viewed close to pole- −0.02 −0.03 −0.08 on, the absorption value derived from the X-ray data appears +0.19 +0.42 +0.92 kT32.97−0.18 2.50−0.40 2.31−0.51 extremely large. To reconcile X-ray and optical/UV absorption +0.72 +0.63 +0.64 EM11.41− 1.46− 1.55− columns requires an optically transparent medium with X-ray 0.36 0.50 0.41 absorption, an issue we will return to in the next section. EM22.73+0.18 2.94+0.27 1.49+0.13 −0.16 −0.20 −0.42 We find no strong overall metal depletion, however, some +0.32 +0.36 +0.31 ff EM38.72−0.32 2.94−0.33 1.37−0.18 di erences for individual abundances are present. While the modelled abundances of medium FIP (O, Si, S) elements are χ2 (d.o.f.) 1.03 (457) 1.02 (147) 0.90 (112) red moderately subsolar, the low FIP (Fe, Mg) elements show a LX 13.2 (20.9) 6.8 (12.5) 3.4 (7.1) higher abundance around or slightly above solar value and especially the high FIP elements (Ne) are enhanced. A high Ne abundance is commonly observed in CTTS and in active stars, but a high Fe abundance is rather atypical for a young CTTS. emission since the OM UVW1 is quite sensitive to flux emitted Depending on the specific model we find a Fe/O ratio of 1–1.5 from the accretion spot region. and Ne/O of 2–2.5. Recently, Telleschi et al. (2007) showed that a dependence on spectral type might be present, with G-type / 3.2. Global spectral properties CTTS showing a higher Fe abundance (or Fe O ratio) as CTTS of spectral type K–M. RU Lup’s spectral type is usually classi- To study the global spectral properties of the X-ray emission fied as late K and its abundance ratios are intermediate between from RU Lup and investigate its changes we used the EPIC data. their K–M-type and G-type population, in contrast to their find- In Fig. 2 we show the PN spectra and the best fit models for the ings. We note that TW Hya, the prototypical accretion dominated three observations to indicate data quality and spectral changes CTTS and also of spectral type late K, shows likewise an Fe/O between the observations. Obviously the spectrum is harder with ratio of 1–1.5. Thus spectral type may not be the only relevant increasing X-ray brightness and the plasma at hot temperatures parameter for abundance irregularities and accretion properties, is more prominent. Very hot plasma (40 MK) is also evidenced for example, may have to be considered. by the presence of the 6.7 keV iron line complex in the data from We also checked for cool plasma which could be responsi- August 8. To quantify the spectral changes, we fit the EPIC data ble for a possible UV/X-ray correlation within individual ob- of the individual observations with a three-temperature model servations. Since moderate absorption is present in the X-ray as described above, and present the derived model parameters spectra, the required plasma temperatures have to be suffi- in Table 2. Note that while absorption and abundances do not ciently high to be responsible for any significant change in vary significantly between the exposures, fits of similar quality the X-ray light curve. Utilising the above derived value of NH, can be obtained with a cool component at very low tempera- we find that plasma temperatures of roughly 2.5 MK are suffi- tures (∼1 MK) and a much larger EM, combined with moder- cient, matching the temperature regime where accretion shock ately higher values for the oxygen abundance AO and interstellar plasma is expected. In our 3-T models this corresponds to plasma column density NH. mainly from the cool and partly from the medium temperature J. Robrade and J. H. M. M. Schmitt: X-rays from RU Lupi: accretion and winds in classical T Tauri stars 233

Fig. 3. Spectral region covering the iron Kα line at 6.4 keV and Fe xxv line complex at 6.7 keV, shown with the three temperature fit (solid)and Fig. 4. Fluxed O vii triplet from RGS data (PSF core, rebinned); the with additional fluorescence line (dashed). background level is indicated by a dashed line. component. This cooler plasma is obscured by hotter plasma emission. A strong iron fluorescence line has also been detected during the more active and X-ray brighter first observation, but in other young stellar objects such as Elias29 (Favata et al. becomes dominant especially in the third observation. Therefore 2005b) or in the several COUP sources in Orion (Tsujimoto the plasma properties derived from the spectral fits are consis- et al. 2005). Theses observations indicate the importance of tent with the presence of a correlation between UV brightness X-ray emission for the ionisation and consequently temporal and soft X-ray emission originating from accretion shocks. A evolution of circumstellar disks. While most previous detec- stronger accretion component would result in a softening of the tions are related to strong flaring, Kα emission from Elias29 X-ray spectra, provided that the usually hotter coronal compo- was also observed in its quiescent state, similar to the detection nent is stable. A softening of the X-ray emission with increasing in RU Lup. Both sources also share the exceptional high coro- brightness was marginally detected during an XMM-Newton ob- nal iron abundance, which has been interpreted by Favata et al. servation of TW Hya (Robrade & Schmitt 2006), but the count (2005b) as evidence for X-ray emitting plasma in magnetic flux rate of RU Lup is not sufficient to calculate meaningful hardness tubes connecting the star and the circumstellar disk. Magnetic ratios and investigate this possibility. field lines between the star and its disk are a natural ingredient of the magnetospheric accretion model and could provide a flow of fresh disk material into the corona. 3.3. Fluorescent emission

In the spectrum of the observation on August 08 some excess 3.4. Oxygen lines – plasma density emission at energies below the 6.7 keV iron line complex is vis- ible, which might be caused by fluorescence emission from iron We used high-resolution RGS spectra to measure emission lines vii viii Kα. After inspection of the PN spectrum in the energy range from O and O , i.e. the resonance (r), intercombination(i) 4.5–8.0 keV we added a narrow (10 eV) Gaussian component at and forbidden ( f ) lines in the He-like triplet of O vii (21.6, 21.8, 6.4 keV to the three temperature model as shown in Fig. 3. We 22.1 Å) and the Ly α line of O viii at 18.97 Å . We searched for find that the quality of the spectral fit to improves significantly cool excess plasma in CTTS compared to main-sequence stars 2 = 2 = viii/ vii from χred 1.33 (11 d.o.f.) to χred 1.03 (10 d.o.f.). The de- via the O O (r) line ratio, presented in Sect. 4 and studied rived flux from the improved spectral model in the fluorescence the density of the cool (2–3 MK) plasma with the f /i- ratio in the line is 5.5±4.3×10−7 photons cm−2 s−1 and the fluorescence line He-like triplet of O vii. A density analysis of moderately hot- from iron Kα at 6.4 keV is present at high significance (>90%). ter plasma at ∼4 MK could be carried out with the Ne ix triplet, Under the assumption that the fluorescence emission from but the unfortunately only moderate SNR of the RGS spec- iron Kα is produced by the illumination of cooler material with tra and non-negligible contamination from the second source X-ray photons – in principle also electron excitation is possi- (RX J1556.4-3749) in this spectral range prevented this analy- ble – the exciting photons need to have energies above 7.11 keV sis. On the other hand, as shown in Fig. 4, the O vii triplet is and the illuminated material must be at temperatures of 2MK. clearly detected and shows an intercombination line significantly While no large flare is present to produce these photons, the stronger than the forbidden line. We extracted the line counts hot plasma component might be sufficiently bright to illumi- from the spectra obtained with the two extraction methods and nate the star and the disk. Additionally the pole-on geometry of list them in Table 3. The derived physical properties were then RU Lup is considered optimal for the production of fluorescence calculated with line intensities corrected for effectiveareaand emission. The required fluorescence efficiency is of at least 5%, absorption as adopted from the EPIC model. The plasma density while for the best fit value up to 15–20% efficiency is necessary. was determined from the relation f / i = R0 / (1 + φ/φc + ne/Nc) These values are somewhat larger than the typical efficiencies of with f and i being the respective line intensities, R0 = 3.95 is 3% derived from Monte Carlo calculations (Bai 1979) to model the low density limit of the line ratio, Nc the critical density and iron fluorescence emission seen during solar flares. However, φ/φc the radiation term, which is neglected in our calculations as on the Sun only the photosphere is illuminated by the X-rays. discussed below. The applied values were taken from Pradhan & Larger efficiencies can be obtained from hidden X-ray sources, Shull (1981). i.e. active regions on the far side of the star and/or with addi- We find a low f / i-ratio independent of the PSF fraction tional target material. A natural explanation would be the sur- used, While actually the f-line is only marginally detected, for- rounding disk, which is likewise illuminated by the stellar X-ray mally the derived values are f / i = 0.26 ± 0.23 for the PSF core 234 J. Robrade and J. H. M. M. Schmitt: X-rays from RU Lupi: accretion and winds in classical T Tauri stars

Table 3. Measured line counts from RGS data for RU Lup for the two extraction regions described in Sect. 2.

OVIII OVII - RGS1 PSF RGS1 RGS2 rif Std. 28.7 ± 8.133.0 ± 8.122.7 ± 7.320.1 ± 6.98.3 ± 5.6 Core 23.8 ± 6.426.3 ± 6.222.9 ± 6.115.3 ± 4.04.0 ± 3.2

and f / i = 0.41±0.31 for the standard region. In both cases, the O vii f / i-ratio is substantially below unity, resulting in a density 11 −3 of ne = 4.4(2.7)×10 cm for the two extraction methods. The lower limit for the density is ∼2(1.5) × 1011 cm−3, on the other hand much higher densities cannot be excluded due to the weak- Fig. 5. Theoretical O viii(Lyα)/O vii(r) line flux ratio calculated with ness of the forbidden line. The PSF core result is closer to the the Chianti database vs. temperature. theoretically expected g-ratio (g = ( f + i)/r) of around unity, but both methods agree within errors. Thus RU Lup is another CTTS that shows a high density in its cool (∼2 MK) plasma component. Two effects, namely the presence of an UV field and of coronal plasma at lower densities, may alter the actual density of the accretion shock plasma in either direction. The strength of the UV field in the proximity of the X-ray emitting plasma is unknown, but its presence may influence the inferred plasma density. A strong UV radiation field would lower the derived plasma densities, however strong UV radiation also has to be attributed to accretion shocks since the photosphere of a late-type star does not produce a sufficiently strong UV flux. On the other hand, a coronal contribution to the cool plasma, which is also present, results in an underprediction of the derived density for the X-ray emitting accretion shock plasma. Therefore we conclude that the bulk of the cool X-ray emitting plasma in RU Lup is generated in accretion shocks.

4. An X-ray view on accretion and winds in CTTS Fig. 6. Ratio of the emitted O viii(Lyα)/O vii(r) line flux vs. oxygen vii + viii RU Lup is a high accretion rate CTTS and it is useful to com- luminosity (O (r) O (Lyα)) for main-sequence stars (diamonds), CTTS (triangles) and the Herbig Ae star AB Aur (square). pare its properties with those of other X-ray bright CTTS. With the increasing number of available medium and especially high- resolution X-ray spectra for these objects we can study signa- (Dere et al. 1997; Landi et al. 2006); it is very sensitive in the tures of two important processes in CTTS and in star-formation temperature range of interest and therefore a powerful diagnostic at large, accretion and outflows. tool. The emissivity curve of O vii(r) dominates over the corresponding O viii(Lyα) curve at temperatures below ∼2.5 MK, and 4.1. Accretion, densities and cool excess their ratio is therefore ideal to investigate cool excess plasma as expected from accretion shock models. While the X-ray emission of CTTS at higher energies (E Specifically, we compared the ratios of CTTS and other ac- 1 keV) is dominated by magnetic activity, density analysis with creting stars with sufficient signal in their high-resolution X-ray He-like triplets and other line diagnostics suggest that accretion spectra to those measured in a large sample of main-sequence processes contribute significantly to the soft X-ray emission in stars at various activity levels (Ness et al. 2004). Note that this many, and maybe even all CTTS. Since the disk truncation radii sample contains several prominent non-accreting young stars are usually large, the accreting material is infalling with almost such as AB Dor, AT Mic and AU Mic. Results for the Herbig Ae free fall velocities and hence post-shock temperatures of up to a star AB Aur were taken from Telleschi et al. (2007), the O vii(r) few MK are unavoidable. Such plasma is still relatively cool with line was inferred from their triplet flux assuming a g-ratio of respect to typical coronal temperatures of active stars and should unity, i.e. (r = f + i). For the CTTS sample we adopted the val- therefore lead to “cool excess” emission. This opens the possi- ues for BP Tau, CR Cha and TW Hya from Robrade & Schmitt bility to investigate accretion scenarios via cool excess emission (2006) with a g-ratio of one for CR Cha due to lower SNR, by comparing accreting stars with stars where only coronal pro- for MP Mus from Argiroffi et al. (2007), for V 4046 Sgr from cesses contribute to the X-ray emission. (Günther et al. 2006) with a distance of 80 pc. T Tau is re- Instead of using broad band fluxes we use the ratio of two analysed here in the same fashion as RU Lup (this work), where strong lines which are commonly observed in stellar X-ray both extraction methods are shown. All given values refer to the spectra, the O viii Lyα line (18.97 Å) and the O vii resonance emitted line fluxes, i.e. they are corrected for the respective ab- line (21.6 Å) with peak formation temperatures of ∼3MKand sorption column. ∼2 MK respectively. In Fig. 5 we show the O viii(Lyα)/O vii(r) In Fig. 6 we plot the absorption corrected O viii line ratio as calculated with the CHIANTI V 5.2 code (Lyα)/O vii(r) line ratio vs. the total oxygen luminosity (the J. Robrade and J. H. M. M. Schmitt: X-rays from RU Lupi: accretion and winds in classical T Tauri stars 235 power emitted in the O viii(Lyα)andOvii(r) lines) for accret- Table 4. Properties of accreting stars and their cool plasma. Masses and ing stars and main-sequence stars. The graph clearly shows that radii from the references in this section and therein, M−1/2R−3/2 in solar our sample CTTS and the main-sequence stars occupy quite dif- units, TS = 3.44 × M/R (shock temp.), Cor.C.= coronal contribution. ferent regions and form two very well separated groups in the (O viii/O vii)vs.(Oviii+O vii) diagram. The correlation be- Star MRM−1/2 R−3/2 Density TS Cor.C. Cool Ex. 11 −3 tween the O viii/O vii line ratio and Loxy for main-sequence stars M R 10 cm MK (EM1/EM2) is well known and caused by the on average higher coronal tem- TW Hya 0.7 1.0 1.20 vhigh 21. 2.4 weak(16.) mod. peratures in the more active and X-ray brighter stars. RU Lup 0.8 1.7 0.50 high 4.4 1.6 mod.(0.6) strong The important point now is that the inferred ratios of all BP Tau 0.8 2.0 0.40 high 3.2 1.4 mod.(0.7) mod. accreting stars lie below the corresponding values for main- MP Mus 1.2 1.3 0.62 high 5.0 3.2 mod.(0.5) weak sequence stars at the same oxygen luminosity. In other words, T Tau 2.4 3.6 0.09 low 0.1 2.3 weak(2.1) strong AB Aur 2.7 2.3 0.17 low 0.1 4.0 mod.(0.6) weak at a given oxygen luminosity an accreting star has a smaller O viii/O vii line ratio than a non-accreting star. Since this ratio depends only on temperature, there must be additional cool plasma radiating predominantly in the O vii lines in accreting stars, i.e. an cool excess.. We emphasize that the above result 4.2. Accretion models vs. observations is robust; investigating the O viii and O vii flux alone we find that the respective O viii flux is compatible or slightly enhanced, We then compared our results with expected accretion shock whereas the O vii flux is significantly enhanced in the accreting parameters from calculations based on the magnetically fun- stars. While most sample CTTS have comparable oxygen lumi- nelled accretion model described in Calvet & Gullbring (1998). nosities of 1–2 × 1029 erg s−1, corresponding to roughly 5–10% Discarding two object from the discussion, V 4046 Sgr (K5+K7, ∼ of their total LX emitted in these two lines, no correlation be- age 6 Myr, Quast et al. 2000) since it is a binary and the dis- tween cool excess and oxygen luminosity is present in our sam- tribution of its X-ray emission among the binary components is ple of accreting stars. Thus the two groups again show different unknown and CR Cha (K2, age ∼1 Myr) since its stellar parame- trends, suggesting a different origin of the observed emission. ters are only poorly constrained, we summarise the stellar prop- In the magnetospheric accretion model a larger cool excess erties of the accreting stars and their fitting results in Table 4. corresponds primarily to a higher accretion luminosity. In addi- Filling factors as derived by Calvet & Gullbring (1998) are in tion, the precise strength of the cool excess depends on the intrin- the range of f = 0.01 for magnetically funneled accretion on sic temperature of the accretion shock plasma and the respective CTTS, e.g. f = 0.007 for BP Tau, but can be up to an order of coronal contribution to the observed cool plasma, however both magnitude smaller or larger in individual stars. Adopting their are affected by interdependencies and uncertainties that arise formulae and assuming free-fall velocity, i.e. the disk truncation from modelling. Besides accretion, other mechanisms have been radius to be much larger than the stellar radius, the infall (post- = × ∼ × −1 × −1/2 −3/2 proposed to produce an excess in the soft X-ray regime, for ex- shock np 4 ni) density is ni M˙ f (M R ). ample disk and coronal stripping of the outer and hottest parts of The accretion luminosity is LAcc ∼ M˙ × M/R, the corresponding the corona (Jardine et al. 2006), disruption of hot plasma struc- plasma temperature is T ∼ V2 ∼ M/R and therefore assumed tures through accreted material (Audard et al. 2005) or filling to be independent of funnelling. Note that different shock tem- of coronal structures with cool material (Preibisch et al. 2005). peratures also influence the observed O viii/O vii ratio and some These mechanisms may likewise be present in young stellar ob- uncertainties on the stellar parameters are also present for the in- jects, however, combining these finding with the often observed vestigated stars. The coronal contribution to the X-ray emission high densities in the cool plasma of CTTS, an accretion shock from the cool plasma is approximated by the ratio EM1/EM2in leads to the required additional, persistent, cool, high density the respective 3-T models, in our sample stars the correspond- plasma component, in a completely natural way. ing temperatures are 2–3 MK (T1) and 6–7 MK (T2). Assuming Note that not all accreting stars generate significant X-ray similar shaped underlying stellar coronal EMDs as determined emission in a high density environment via magnetically fun- from active stars, this ratio traces the importance of the coro- nelled accretion streams. While the exact density of the accre- nal contribution to the cool plasma. We denote EM1/EM2 ratios tion plasma depends on the presence of UV-fields and the coro- above 2 as “weak”, in the range 0.5–2 as “moderate” and below nal contribution to the cool plasma, all analysed CTTS with a 0.5 as “strong”. Finally we roughly quantified the strength of the moderate oxygen luminosity exhibit cool plasma at high densi- cool excess by comparing the O viii(Lyα)/O vii(r) line ratio for 11 −3 ties (ne 10 cm ). This requires a strong funnelling of the ac- the coronal sources that exhibit the lowest value at a given oxy- cretion stream, independent of the strength of their cool excess. gen luminosity and the respective accreting sources. We denote Contrary, the star with the largest cool excess and especially the a cool excess as “weak” for a ratio 2, “moderate” for a ratio of largest oxygen luminosity, T Tau itself, is the only T Tauri star ∼2–4 and “strong” for larger ratios. that exhibits cool plasma at a lower density and in fact stars like Altogether the characteristics of the accreted plasma depend T Tau are often not classified as CTTS, but as intermediate mass on an interplay of stellar properties and the accretion stream. TTS (IMTTS) that are thought to be predecessors or analogs of The dependence on stellar properties clearly favours the high Herbig AeBe stars. In a related analysis presented by Telleschi density scenario for more compact low-mass stars, and predicts et al. (2007), that studies a TTS sample in Taurus-Auriga and lower plasma temperatures and therefore larger soft excess for shows a soft excess in accreting stars compared to WTTS, the less contracted objects as long as X-ray temperatures are reached Herbig Ae star AB Aur is found as another example of an accret- in the accretion shock. Considering the accretion stream, the ra- ing star that exhibits a cool excess and a low density cool plasma. tio M˙ / f is important for the resulting density, i.e. large mass ac- Both stars that show no high density plasma are of higher mass cretion rates and small filling factors are favoured in the higher (2 M) than the sample CTTS (1 M), suggesting a scenario density regime. A strong cool excess is produced by large mass where accretion properties are linked to stellar properties, i.e. accretion rates via a large accretion luminosity. The accretion mass and radius. luminosity also increases for more compact objects, but when 236 J. Robrade and J. H. M. M. Schmitt: X-rays from RU Lupi: accretion and winds in classical T Tauri stars considering a cool excess as measured by O viii/O vii it con- have only a minor impact on the cool excess at least in CTTS. flicts with the corresponding higher shock temperatures. We note that CR Cha and V 4046 Sgr mostly fit into this pic- There is a remarkable correlation between the cool plasma ture since they are K-type low-mass stars and also show the ex- properties as expected in the magnetospheric accretion model pected high plasma density. Admittedly our sample is not very from stellar parameters alone and the observed properties for large and completely lacking of any CTTS with very low mass the six investigated stars. Notably the measured densities cor- (0.5 M) due to the fact that they are usually X-ray darker since −1/2 −3/2 relate well with the value of the quantity M R suggesting LX correlates with mass in T Tauri stars. If also applicable in this that the ratio M˙ / f does not vary extremely between the sample mass regime, we expect early M-type CTTS to show rather high CTTS and that the cool plasma is indeed dominated by the accre- plasma densities that should be observable in X-rays, if accretion process in most cases. The property M−1/2R−3/2 describes tion rates are high enough and the accretion plasma is heated to on the one hand the compactness of a star for a given mass but sufficiently high temperatures. also some kind of general smallness. The only star with very One parameter, i.e. stellar age, is completely absent in the 12 −2 high density (ne > 10 cm ), TW Hya, is also the only star above discussion. The stellar evolution is of course “hidden" with (M−1/2R−3/2) > 1, all stars with 0.4 (M−1/2R−3/2) 0.6show in the stellar radius for a given mass and consequently old 11 −2 high densities (ne > 10 cm ) and the other two stars with ( 10 My) and more compact CTTS like TW Hya or MP Mus 10 −2 −1/2 −3/2 low densities (ne ≤ 10 cm )have(M R ) < 0.2. While exhibit higher shock temperatures than young ( 1 My) CTTS among the high density stars the correlation is roughly linear, such as BP Tau. Likewise the mass accretion rates of CTTS de- the larger variations in density compared to those of M−1/2R−3/2 crease on average throughout their evolution, further diminish- for the extreme cases of very high and low density suggest that ing the cool excess with time. While there is probably a large an additional correlation with M˙ / f might amplify this trend. spread, this trend is also present in the observations and none of Consequently the funnelling would have to be very effective the older CTTS, TW Hya, MP Mus and V 4046 Sgr, exhibit any for compact and tiny objects and much less pronounced for ex- strong cool excess. On the other hand, all older CTTS show quite panded or more massive stars. Note, that the density of the cool high plasma densities suggesting moderate mass accretion rates plasma component does not correlate with either the shock tem- and a well funneled accretion stream. This proposal of a uni- perature or with the cool excess of the respective star. fied scenario for the observed X-ray properties of accreting stars The temperature of the accreted plasma is also determined needs to be tested with a larger sample to improve the statistics by stellar mass and radius, but for the strength of the cool ex- especially among the very low and intermediate mass regime, cess the coronal contribution and its X-ray brightness relative to but provides a plausible explanation for the similarities as well the one of the accretion spot is more important, since our defini- as the differences in the observed characteristics of X-ray bright tion of a cool excess involves O viii emission which traces hot- accreting stars. ter plasma. Necessarily the intrinsic temperature of the shocked / ffi plasma, which depends on the ratio M R, has to be su ciently 4.3. Outflows and anomalous X-ray absorption low to produce a cool excess that is traced by enhanced O vii emission. The calculated shock temperatures are in the range In the X-ray spectra of the almost pole-on RU Lup we measure between 1.5–4.0 MK, i.e. the temperature range where the ioni- an absorbing column density clearly incompatible with optical sation stage changes from a dominating O vii contribution to a or UV measurements. In Fig. 7 we show the X-ray spectrum of dominating O viii contribution in a collisionally ionised plasma. RU Lup as observed on August 08 with our best fit model and In two stars, MP Mus and AB Aur, the calculated shock tem- also with the best fit model using the absorption value derived perature is already above 3 MK and their cool excess is indeed from UV measurements Herczeg et al. (2005). Column densi- weakest in our sample. That the Herbig AeBe star AB Aur shows ties derived from optical observations are usually comparable or some cool excess at all is surprising given its shock temperature even lower. Besides the fact that the applied model is quite un- of 4 MK. This might be explained by inhomogeneous accretion physical since it does not contain any cool plasma, it still cannot spots that arise from recent modelling (Romanova et al. 2004). explain the observed spectrum. Therefore we have to attribute These spots show for example an intrinsic temperature distri- the observed discrepancy to the presence of an unknown X-ray bution, an effect that might be more pronounced especially for absorbing material along the line of sight. The same effect was less funneled accretion streams. A strong cool excess is only ob- also detected in several other CTTS such as AA Tau (Schmitt & served in stars with shock temperatures below 2.5 MK, but stars Robrade 2007), a CTTS viewed under an intermediate inclina- with similar shock temperatures may show quite different other tion of ∼75◦ (Bouvier et al. 1999) and in the also pole-on CTTS accretion properties as discussed exemplarily for the extreme TW Hya (Robrade & Schmitt 2006), where the effect is an or- cases T Tau and TW Hya. In T Tau a large mass accretion rate of der of magnitude lower and the discrepancy could be instead −8 3–6 ×10 M/yr (Calvet et al. 2004), moderate shock temper- attributed to modelling or calibration uncertainties. atures and a weak coronal contribution are present, producing Reconciliation of the X-ray and UV/optical absorption ob- the largest cool excess in our sample; its lower plasma density servations requires an optically transparent, but sufficiently requires a large filling factor and/or possibly involves other X-ray absorbing medium. This medium needs to be free of any “cooling mechanisms” like mass loading of coronal structures as significant contribution of larger dust grains but has to contain suggested by Güdel et al. (2007). On the other hand, TW Hya, a lot of X-ray absorbing material such as gas. Additionally the the exceptional case of an accretion dominated CTTS shows absorption component has to be persistent and relatively steady no extreme cool excess. However, its shock temperature is over at least a month. A good candidate for such a medium could only average and in addition the lower mass accretion rate of be a strong outflow either from the star, possibly powered by the −9 ∼10 M/yr (Muzerolle et al. 2000; Alencar & Batalha 2002) accretion process (Matt & Pudritz 2005), and/or from the inner contributes to TW Hya’s only moderate cool excess. Its high disk (Konigl & Pudritz 2000; Alencar et al. 2005). Likewise the plasma density then points to a strong funnelling of the ac- accretion streams may contribute to the additional absorption. creted plasma. Further modifications, for example through dif- Another more exotic possibility could be a dust envelope with a ferent disk truncation radii or levels of stellar activity, seem to very peculiar dust grain distribution containing sufficiently small J. Robrade and J. H. M. M. Schmitt: X-rays from RU Lupi: accretion and winds in classical T Tauri stars 237

the rim of a warped disk which partially occults the star at recurring phases (Bouvier et al. 1999) and associated material, accretion streams or outflows from the disk likely add significantly to its large and variable X-ray absorption (Schmitt & Robrade 2007). We therefore conclude that the available X-ray data can indeed constrain the amount of material along the line of sight and yield – when adopting a simplified wind model with an outflow velocity of a few hundred km s−1 – mass loss rates at the percent level of the mass accretion rates. Interpreting therefore the observed X-ray absorption as arising from a wind thus yields at least a physically consistent picture.

Fig. 7. X-ray spectra of the almost pole-on CTTS RU Lup. The his- tograms show our model (black) and the best fit model with absorption 5. Conclusions set to values derived from the UV measurements (red). From our studies of the X-ray emission from RU Lup and other CTTS we draw the following conclusions: grains. Considering the nearly pole-on view of RU Lup, a wind 1. RU Lup is another example of a CTTS where cool high den- scenario appears natural, in particular, since strong, fast winds sity plasma is present. The density of 3−4 × 1011 cm−3 as are well known for CTTS. However, the origin and geometry of deduced from the O vii triplet supports an accretion shock the wind strongly depend on the underlying model and detailed scenario for the bulk of cool plasma. This cool plasma com- properties of CTTS winds are the subject of considerable debate. ponent is quite stable over a month. In general, a stellar wind requires lower mass outflow rates 2. Spectral variations indicate a change from an EMD that is than a disk wind, since it originates from the stellar surface and clearly dominated by coronal activity in the first observation therefore obscures the X-ray emitting regions more effectively. to a phase with a more equal distribution of cooler and hot- On the other hand, a stellar wind originating from coronal struc- ter plasma in observation three. Many examples of coronal tures in analogy to the solar wind would have temperatures of at dominated CTTS have been detected so far, but RU Lup is a least several 100 000 K, much higher than the few 10 000 K usu- rare example where a change from state dominated by coro- ally attributed to CTTS winds. A hot wind has been claimed on nal activity to a state with a major contributing cool accretion the basis of asymmetric O vi line profiles observed in TW Hya component is observed over a month. Its X-ray luminosity is by Dupree et al. (2005); however, Johns-Krull & Herczeg (2007) not correlated with UV brightness over the campaign, how- question this interpretation and argue for a much cooler wind in ever during X-ray fainter phases there are indications for a TW Hya. Unfortunately our data do not allow us to investigate correlation on timescales of several hours. the properties of the absorbing wind component in more detail. 3. In all investigated accreting stars the characteristics of the If we adopt – as the simplest absorption model – that of a cooler X-ray emitting plasma are influenced by the accre- spherically symmetric wind with constant velocity, one derives tion process. An excess of cool plasma, as evidenced by a a line-of-sight absorption column of NH = n0 × R0 with n0 de- lower O viii(Lyα)/O vii(r) line-ratio, is present in our sample noting the plasma density at the wind base and R0 its distance stars when compared to main-sequence stars. The strength from the star. Assuming the wind base to be essentially at the of the cool excess depends on an interplay of accretion surface of the star and adopting the numbers of RU Lup, i.e. shock luminosity and temperature as well as the omnipresent 21 −2 NH ∼ 1.7 × 10 cm and R0 = 1.7 R,wefindabaseden- coronal contribution. Cool, high density plasma is found 10 −3 sity of 1.4 × 10 cm . Further assuming this outflow to occur so far exclusively in the low-mass CTTS sample 1 M), −1 over 4π steradian with a velocity of 300 km s , results in a mass while accreting stars with intermediate mass (2 M)show −9 loss rate of M˙ wind ∼ 2 × 10 M/yr. However, the otherwise de- lower densities. Many aspects of the accretion process can be rived mass loss rates of CTTS are extremely uncertain, but are explained by stellar mass and radius and their evolution with thought to be in the range 0.01–0.1 of the respective mass accre- time in a qualitative way. We suspect a relation to mass ac- tion rates (Hartigan et al. 1995; Konigl & Pudritz 2000), which cretion rates and especially amount of funnelling, which pro- in turn are also only poorly constrained and in addition time vari- duce the different properties of the accretion shock plasma able. Recent estimates for the mass accretion rates for RU Lup that are seen in the respective X-ray spectra. −8 are about 5 ± 2 × 10 M/yr (Herczeg et al. 2005), and therefore 4. We derive from X-ray spectra an absorption column of −9 21 −2 amasslossrateofM˙ wind ∼ 2 × 10 M/yr would perfectly fit 1.8 × 10 cm for RU Lup, roughly an order of magnitude the “expectations”. Carrying out the same exercise for TW Hya above the value derive from Lyα absorption. To reconcile the 20 −2 (NH ∼ 3×10 cm ) yields again consistent results if one adopts optical extinction and X-ray absorption towards RU Lup, one −9 mass accretion rates of 0.5−2 × 10 M/yr (Muzerolle et al. needs X-ray absorption in an optically almost transparent 2000; Alencar & Batalha 2002). Note that outflow velocity, wind medium. Large discrepancies between absorption values de- solid angle and location of the wind base appear in this model rived from X-ray spectra and in the optical/UV wavelength only as linear parameters and moderate changes of their values regimes are also found in several other CTTS. We suggest do not change the rough estimations of the mass loss rates given that strong outflows/winds are responsible, which originate above. Accordingly, the lower X-ray absorption for TW Hya either from the star and/or the disk. Clearly strong winds would correspond to a more moderate wind, for RU Lup both are present in RU Lup and the nearly pole-on view does not values are roughly by a magnitude larger and may well be ex- favour anomalous dust material to be responsible for the explained in a similar, scaled up scenario. AA Tau is viewed over cess absorption; however, a contribution from the accretion 238 J. Robrade and J. H. M. M. Schmitt: X-rays from RU Lupi: accretion and winds in classical T Tauri stars

ﬂows is expected. Using the required additional X-ray ab- Hartigan, P., Edwards, S., & Ghandour, L. 1995, ApJ, 452, 736 sorption columns and a simple wind model we derive mass Herczeg, G. J., Walter, F. M., Linsky, J. L., et al. 2005, AJ, 129, 2777 loss rates of a few percent of the stellar mass accretion rates Hughes, J., Hartigan, P., & Clampitt, L. 1993, AJ, 105, 571 Jardine, M., Cameron, A. C., Donati, J.-F., Gregory, S. G., & Wood, K. 2006, for CTTS. MNRAS, 367, 917 Johns-Krull, C. M., & Herczeg, G. J. 2007, ApJ, 655, 345 Acknowledgements. This work is based on observations obtained with Kashyap, V., & Drake, J. J. 2000, Bull. Astron. Soc. Ind., 28, 475 XMM-Newton, an ESA science mission with instruments and contributions di- Kastner, J. H., Huenemoerder, D. P., Schulz, N. S., Canizares, C. R., & rectly funded by ESA Member States and the USA (NASA). J.R. acknowledges Weintraub, D. A. 2002, ApJ, 567, 434 support from the DLR under grant 50OR0105. Konigl, A., & Pudritz, R. E. 2000, Protostars and Planets IV, 759 Lamzin, S. A. 2000, Astron. Lett., 26, 225 Lamzin, S. A., Bisnovatyi-Kogan, G. S., Errico, L., et al. 1996, A&A, 306, 877 References Landi, E., Del Zanna, G., Young, P. R., et al. 2006, ApJS, 162, 261 Loiseau, N., Ehle, M., Pollock, A., et al. 2006 Alencar, S. H. P., Basri, G., Hartmann, L., & Calvet, N. 2005, A&A, 440, 595 Matt, S., & Pudritz, R. E. 2005, ApJ, 632, L135 Alencar, S. H. P., & Batalha, C. 2002, ApJ, 571, 378 Muzerolle, J., Calvet, N., Briceño, C., Hartmann, L., & Hillenbrand, L. 2000, Appenzeller, I., Krautter, J., & Jankovics, I. 1983, A&AS, 53, 291 ApJ, 535, L47 Argiroﬃ, C., Maggio, A., & Peres, G. 2007, A&A, 465, L5 Ness, J.-U., & Wichmann, R. 2002, Astron. Nachr., 323, 129 Arnaud, K. A. 1996, in Astronomical Data Analysis Software and Systems V, Ness, J.-U., Güdel, M., Schmitt, J. H. M. M., Audard, M., & Telleschi, A. 2004, ed. G. H. Jacoby, & J. Barnes, ASP Conf. Ser., 101, 17 A&A, 427, 667 Audard, M., Güdel, M., Skinner, S. L., et al. 2005, ApJ, 635, L81 Neuhäuser, R., Sterzik, M. F., Schmitt, J. H. M. M., Wichmann, R., & Krautter, Bai, T. 1979, Sol. Phys., 62, 113 J. 1995, A&A, 297, 391 Bouvier, J., Chelli, A., Allain, S., et al. 1999, A&A, 349, 619 Pradhan, A. K., & Shull, J. M. 1981, ApJ, 249, 821 Calvet, N., & Gullbring, E. 1998, ApJ, 509, 802 Pravdo, S. H., Feigelson, E. D., Garmire, G., et al. 2001, Nature, 413, 708 Calvet, N., Muzerolle, J., Briceño, C., et al. 2004, AJ, 128, 1294 Preibisch, T., Kim, Y.-C., Favata, F., et al. 2005, ApJS, 160, 401 Dere, K. P., Landi, E., Mason, H. E., Monsignori Fossi, B. C., & Young, P. R. Quast, G. R., Torres, C. A. O., de La Reza, R., da Silva, L., & Mayor, M. 2000, 1997, A&AS, 125, 149 in IAU Symposium Dupree, A. K., Brickhouse, N. S., Smith, G. H., & Strader, J. 2005, ApJ, 625, Robrade, J., & Schmitt, J. H. M. M. 2006, A&A, 449, 737 L131 Robrade, J., & Schmitt, J. H. M. M. 2007, A&A, 461, 669 Favata, F., Flaccomio, E., Reale, F., et al. 2005a, ApJS, 160, 469 Romanova, M. M., Ustyugova, G. V., Koldoba, A. V., & Lovelace, R. V. E. 2004, Favata, F., Micela, G., Silva, B., Sciortino, S., & Tsujimoto, M. 2005b, A&A, ApJ, 610, 920 433, 1047 Schmitt, J. H. M. M., & Liefke, C. 2004, A&A, 417, 651 Favata, F., Bonito, R., Micela, G., et al. 2006, A&A, 450, L17 Schmitt, J. H. M. M., & Robrade, J. 2007, A&A, 462, L41 Feigelson, E. D., & Montmerle, T. 1999, ARA&A, 37, 363 Schmitt, J. H. M. M., Robrade, J., Ness, J.-U., Favata, F., & Stelzer, B. 2005, Gahm, G. F. 1980, ApJ, 242, L163 A&A, 432, L35 Grevesse, N., & Sauval, A. J. 1998, Space Sci. Rev., 85, 161 Smith, R. K., Brickhouse, N. S., Liedahl, D. A., & Raymond, J. C. 2001, ApJ, Güdel, M., Skinner, S. L., Briggs, K. R., et al. 2005, ApJ, 626, L53 556, L91 Güdel, M., Skinner, S. L., Mel’Nikov, S. Y., et al. 2007, A&A, 468, 529 Stelzer, B., & Schmitt, J. H. M. M. 2004, A&A, 418, 687 Günther, H. M., Liefke, C., Schmitt, J. H. M. M., Robrade, J., & Ness, J.-U. Stempels, H. C., & Piskunov, N. 2002, A&A, 391, 595 2006, A&A, 459, L29 Telleschi, A., Güdel, M., Briggs, K. R., Audard, M., & Scelsi, L. 2007, A&A, Günther, H. M., Schmitt, J. H. M. M., Robrade, J., & Liefke, C. 2007, A&A, 468, 443 466, 1111 Tsujimoto, M., Feigelson, E. D., Grosso, N., et al. 2005, ApJS, 160, 503